Professional Documents
Culture Documents
Long Distance Mobile MAX DOAS Observations of NO2 and SO2 Over 2020 Atmosphe
Long Distance Mobile MAX DOAS Observations of NO2 and SO2 Over 2020 Atmosphe
Atmospheric Research
journal homepage: www.elsevier.com/locate/atmosres
ABSTRACT
In this study, the spatiotemporal distribution of the NO2 and SO2 Vertical Columns Densities (VCDs) in the North China Plain (NCP) region was measured by long-
distance mobile measurements using a mobile Multi-AXis Differential Optical Absorption Spectroscopy (MAX-DOAS) instrument. The mobile observations were
performed in both summer (July 2017) and winter (January and February 2018) and the total driving mileage exceeded 3000 km. The concentrations of NO2 and SO2
in different seasons and areas were significantly different. During winter, serious NO2 and SO2 pollution was observed in northern Anhui Province, central Shandong
Province, and the Beijing-Tianjin-Hebei Region. The evolution and transportation processes of three typical heavy pollution cases were discussed in detail. Combined
with the Weather Research and Forecasting with Chemistry model simulated wind field information, the NO2 transportation flux from northern Jiangsu Province to
northern Anhui Province is quantified as 7.12 kg s−1. Finally, we estimated the NO2 and SO2 emissions from the Dezhou and Hengshui power plants by plume cross
section scanning and encircled observations, respectively. The NO2 and SO2 emission fluxes of the Dezhou Power Plant are 0.79 kg s−1 and 1.11 kg s−1, respectively,
while the NO2 and SO2 emission fluxes of the Hengshui Power Plant are 0.12 kg s−1 and 0.36 kg s−1, respectively. This study quantitatively analyzed the trans-
portation of atmospheric pollutants and power plant emissions, which is helpful for understanding the occurrence and evolution of pollution and is also useful for the
government to introduce policies to protect and control the atmospheric environment.
1. Introduction total area of cultivated land in China. Furthermore, about 20% of the
total population in China lives in this region. Owing to rapid in-
The North China Plain (NCP) region is bounded by the Yan dustrialization and urbanization, the NCP region is facing severe atmo-
Mountains (north), Dabie Mountains and Yangtze River (south), Taihang spheric pollution problems, especially the Beijing-Tianjin-Hebei region. It
Mountains (west), and Bohai and Huanghai seas (east). This region usually suffers from severe haze pollution combined with high con-
contains the major parts of Hebei, Henan, and Shandong Provinces and centrations NO2 and SO2 pollution events during winter. Moreover, the
the northern parts of Anhui and Jiangsu Provinces. The megacities of heavy regional pollution has spread to an even larger area, lasted for a
Beijing and Tianjin are also included in this region. Although the land long time, and seriously impacted the lifestyle and health of residents
area of NCP region only takes up 3.22% of the total China land area, but (Zhao et al., 2013; Fu et al., 2014; Sun et al., 2014; Wu et al., 2017a; Su,
the area of cultivated land in this region accounts for about 20% of the 2018; Zhang et al., 2018; Wang et al., 2019).
⁎
Correspondence to: C. Liu, Department of Precision Machinery and Precision Instrumentation, University of Science and Technology of China, Hefei 230026,
China.
⁎⁎
Correspondence to: S. Wang, Shanghai Key Laboratory of Atmospheric Particle Pollution and Prevention (LAP3), Department of Environmental Science and
Engineering, Fudan University, Shanghai 200433, China.
E-mail addresses: chliu81@ustc.edu.cn (C. Liu), shanshanwang@fudan.edu.cn (S. Wang).
https://doi.org/10.1016/j.atmosres.2020.105037
Received 24 December 2019; Received in revised form 13 April 2020
Available online 16 May 2020
0169-8095/ © 2020 Published by Elsevier B.V.
W. Tan, et al. Atmospheric Research 245 (2020) 105037
Table 1
The detail information of mobile measurement operations.
Date Colour of route Observation route Observation time (LT) Observation distance (km)
* The observations under the solar zenith angle (SZA) larger than 80° are marked in brackets, and the different cities are represented by abbreviated capital letters
(BJ: Beijing, TJ: Tianjin, LF: Langfang, CZ: Cangzhou, BD: Baoding, HS: Hengshui, DZ: Deizhou, JN: Jinan, TA: Taian, JNI: Jining, ZZ: Zaozhuang, XZ: Xuzhou, HB:
Huaibei, SZ: Suzhou, BB: Bengbu, HN: Huainan, HF: Hefei).
The sources emissions, regional pollutant transport, chemical pro- the pollution characteristics and related transportation processes in the
cesses, and unfavorable meteorological conditions for diffusion have NCP region in summer (August 2017) and winter (January and
been identified as the most important causes of the heavy pollution February 2018). The long-distance mobile MAX-DOAS observations
events (Chan, 2017). Owing to rapid development and dense anthro- revealed the spatial distribution characteristics of the NO2 and SO2
pogenic activities, there are many industrial, power, transport, and tropospheric vertical column densities (VCDs) over large areas of the
residential emissions in the NCP region (Xing et al., 2011; Zhang et al., NCP region. In addition, we also quantitatively analyzed the typical
2018; Chen et al., 2019). Therefore, the government has recently for- pollutant transportation from northern Jiangsu Province to northern
mulated environmental protection policies, such as adjusting the in- Anhui Province during the pollution cases. Finally, the emissions of NO2
dustrial structure, shutting down polluting factories, restricting vehicle and SO2 from the typical point sources of the Hengshui and Dezhou
use through license plate rules during heavy pollution periods, and power plants were estimated.
other integrated regional management measures (MEP, 2017). Mean-
while, the regional transportation of pollutants is significantly influ- 2. Methodology
enced by meteorological conditions such as the wind speed, wind di-
rection, atmospheric temperature and pressure conditions. The static 2.1. Mobile MAX-DOAS measurements
and stable meteorological conditions can easily cause the accumulation
of atmospheric pollutants, and are not conducive to the diffusion and 2.1.1. Routes of mobile MAX-DOAS observations
transportation of pollutants. However, the atmospheric pollutants can In this campaign, mobile observations were conducted over a total
be also transported by horizontal advection, which may cause atmo- of 9 days, including 2 days in summer (August 2017) and 7 days in
spheric pollution over larger areas (Xu et al., 2011; Jin et al., 2016; Wu winter (January and February 2018). The details of the daily mea-
et al., 2018). Moreover, atmospheric chemical processes are also a surement routes and distances are listed in Table 1. As shown in Fig. 1,
major reason for secondary aerosol, O3, and some other pollution. With the mobile measurements crossed large parts of the NCP region, and the
the favorable atmosphere conditions for chemical reactions, the im- total driving mileage exceeded 3000 km. The distance from Hefei at the
portant precursors of NO2 and SO2 tend to form nitrate and sulfate, and southernmost end to Beijing at the northernmost end is about 1000 km.
finally form secondary aerosols (Jang and Kamens, 2001; Huang et al., These routes pass through more than 17 cities. Fig. 1(a), (b), and (c)
2014). present the three completed long-distance north to south observations,
With the advantages of spatial and temporal coverage, satellite which were implemented on 21 to 22 August 2017, 15 to 17 January
observations are widely used to investigate large-scale pollution pro- and 09 to 10 February 2018, respectively.
cesses and long-distance pollutant transport cases (Tao et al., 2012;
Wang et al., 2012; Zhang et al., 2018; Zhang et al., 2019). However, this 2.1.2. Experimental setup
method does not have sufficient sensitivity to the surface layer (lower During the summer and winter observations, two different designed
troposphere), where the main atmospheric pollutants are located. As a MAX-DOAS instruments were used. For the summer observations a
detection method with high sensitivity and continuity, Multi-AXis Dif- MAX-DOAS instrument combined with a GPS recorder were applied on
ferential Optical Absorption Spectroscopy (MAX-DOAS) technique has a passenger car, as described in (Tan et al., 2019). A compact Airyx
been widely employed in recent years. As one of the applications of the MAX-DOAS instrument which was special design for mobile observa-
DOAS method (Platt and Stutz, 2008), it is widely used to retrieve the tion has been used in winter observations. The compact MAX-DOAS
vertical distribution and total columns of trace gases, e.g., NO2, SO2, system has integrated a GPS recorder, a laptop computer and an MAX-
HCHO, HONO, and aerosols (Clémer et al., 2010; Irie et al., 2011; DOAS instrument. This system has been used in an ship-based study in
Pinardi et al., 2013; Wang et al., 2014; Chan et al., 2015; Chan et al., the East China Sea (Tan et al., 2018), and similar instruments have also
2017; Chan et al., 2019). In addition, some MAX-DOAS studies have been used in relevant studies (Kreher et al., 2019; Wang et al., 2020).
also evaluated trace gas horizontal distribution in recent years (Ortega Both the two MAX-DOAS instruments consist of a spectrometer, a
et al., 2015; Casaballe et al., 2017; Seyler et al., 2019; Casaballe et al., temperature controlling unit, a scanning telescope, a connecting fiber,
2020). Owing to its portability, the MAX-DOAS instrument can be controlling electronic devices, and other necessary devices. The spec-
carried on mobile platforms to obtain the spatiotemporal distribution of trometers of both them are Avantes spectrometer (AvaSpec-ULS2048L-
atmospheric pollutants over large areas and obtain the emission char- USB2) cover the same wavelength range of 300 nm to 460 nm with a
acteristics by encircling or scanning the pollution sources (Ibrahim Full Width Half Maximum (FWHM) spectral resolution of 0.6 nm, with
et al., 2010; Shaiganfar et al., 2011; Shaiganfar et al., 2017; Tan et al., a one-dimensional CCD detector (Sony ILX511, 2048 individual pixels).
2019). However, compared with the former instrument the latter one opti-
In this study, a MAX-DOAS instrument was on board a passenger mized the performance of some configurations, such as the field of view
car, and used to measure the tropospheric VCD of NO2 and SO2 over (FOV) of telescope changed from 0.5 to 0.3 and the type of optical fiber
NCP region. Based on the mobile MAX-DOAS observations, we studied changed from single-core fiber to multi-core optical fiber. The signal-to-
2
W. Tan, et al. Atmospheric Research 245 (2020) 105037
Fig. 1. The routes of mobile MAX-DOAS measurements, which are indicated by different colors for the observations during different days. (a) shows the summer
observations in August 2017, while (b) and (c) are the winter observations in January and February 2018.
noise ratio (SNR) of the latter instrument is better than that of the (Meller and Moortgat, 2000), O3 at 223 K and 243 K (Serdyuchenko
former, especially in short wave radiation. For the summer measure- et al., 2014), and BrO at 223 K (Fleischmann and Hartmann, 2004)
ments in August 2017, the measurement sequence of elevation angles combined with a Ring spectrum were included in the DOAS fitting
(α) of 1 × 90° and 4 × 30° was chosen. With the temporal resolution of process. Different to the retrieval of NO2, we included the trace gas
30 s for individual spectrum measurement, a full scanning sequence absorption cross section of SO2 at 298 K (Vandaele et al., 2009) and
took about 3 min in total. During the winter observations from 15 to 17 removed the cross sections of O4 at 293 K and NO2 at 220 K for the SO2
January 2018, the elevation angle scanning sequence was set to be analysis. A 5-order polynomial was included in the DOAS fitting for the
1 × 90° and 3 × 30°. An individual spectrum measurement took about two trace gases. Moreover, the wavelength calibration was achieved by
20 s, and the whole scanning sequence duration about 1.5 min. To fitting the measured zenith spectrum to a high resolution solar spec-
achieve the higher temporal resolution of zenith measurements, the trum (Chance and Kurucz, 2010). These NO2 retrieval settings have
elevation scanning sequence was changed to the combination of been used during Second Cabauw Intercomparison campaign for Ni-
1 × 90° and 1 × 30° from 24 January 2018. The duration of an in- trogen Dioxide measuring Instruments (CINDI-2) campaign (Kreher
dividual spectrum measurement was about 30 s, while each scanning et al., 2019), while the SO2 retrieval settings also referred to the pre-
sequence took about 1 min. All of these elevation angle settings include vious relevant study (Theys et al., 2015).
zenith (90°) and low elevation angle (30°) observations. At first the As the measured zenith spectrum of each elevation angle sequence
mobile observations were performed along the expressway, which far was selected as the Fraunhofer reference spectrum, the DOAS retrieval
away from some larger emission sources, the zenith observations were yielded the Differential Slant Column Densities (DSCDs). A typical
set up in low frequency. However, the frequency of zenith measure- DOAS spectral fitting process of the spectrum measured at the elevation
ments has increased during the observations close to important emis- of 30° at 13:59 local time (LT) on 15 January 2018 is presented in
sion sources such as cities and power plants. The daily measured scat- Fig. 2. The retrieved DSCDs of NO2 and SO2 are 3.18 × 1016 molec
tered sunlight spectra were automatically collected in a laptop cm−2 and 3.52 × 1016 molec cm−2, respectively. The clear absorption
computer. In addition, a high-precision Global Position System (GPS) structure and relatively low residuals indicate the good performance of
data receiver was used to record the real-time coordinate positions spectral fitting. Owing to the influenced by sunlight blocked by build-
information of the mobile measurements. ings and other objects, there are some failure fitting results. For the NO2
DSCDs under SZA < 80° and the DOAS fit root mean square
(RMS) < 0.003 were retained, while the SO2 DSCDs under SZA < 70°
2.1.3. DOAS retrieval and the RMS < 0.005 were included in the discussion. Afterwards,
The measured spectra were analyzed through the QDOAS spectral there are about 88% and 64% DSCD results remains for NO2 and SO2,
fitting software suite, which is based on the DOAS technique and de- respectively.
veloped by BIRA-IASB (http://uv-vis.aeronomie.be/software/QDOAS/
). Before the DOAS fitting, the measured spectra were first corrected by
2.1.4. Retrieval vertical column densities of trace gases
the offset and dark current. Furthermore, the zenith measurements of
To obtain the VCD results, the DSCDs were converted via the Air
each sequence were used as the Fraunhofer reference spectrum. The
Mass Factors (AMFs) using the following formula:
details of the retrieval settings are listed in Table 2. The wavelength
ranges of 338 nm to 370 nm and 312 nm to 326 nm were chosen for SCDtrop DSCD
NO2 and SO2 retrieval, respectively. For the analysis of NO2, the trace VCDtrop = =
AMFtrop AMF 90
°
AMF = 90
° (1)
gas absorption cross sections of O4 at 293 K (Thalman and Volkamer,
2013), NO2 at 220 K and 298 K (Vandaele et al., 1998), HCHO at 297 K where α is the elevation angle of the measurement. Besides the zenith
3
W. Tan, et al. Atmospheric Research 245 (2020) 105037
Table 2
DOAS spectral fitting settings of NO2 and SO2.
Parameter Data source Trace gases
NO2 SO2
observations, there were only 30° measurements during the mobile et al., 2018). In addition, this product has been validated to be more
campaign. Thus, the AMF can be calculated easily and quickly through representative over China because more accurate NO2 a priori profiles
the geometric approximation method, which is a widely used method were used in the AMF calculation (Liu et al., 2016; Su et al., 2017; Xing
that has been validated by the radiative transfer model (RTM) calcu- et al., 2017; Hong et al., 2018; Tan et al., 2018).
lation (Hönninger et al., 2004; Brinksma et al., 2008; Wagner et al.,
2010; Halla et al., 2011; Tan et al., 2018).
2.3. Estimation of pollutant transportation and source emissions
AMF = 1 sin( ) (2)
In this study, mobile observations were conducted to estimate the
As calculated by Eq. (2), the AMF of the 30° elevation angle is 2 pollutant transportation and source emissions over the NCP region.
while the AMF of the 90° elevation angle is 1; thus, the tropospheric Combined with the wind information, the pollutant transportation and
VCDs of the trace gases are approximately equal to the DSCDs for the source emission fluxes can be acquired through the integral along the
measurements under the elevation angle of 30°. driving route using the following equation (Ibrahim et al., 2010):
The Ozone Monitoring Instrument (OMI) was onboard the NASA where VCD indicates the tropospheric VCD of trace gases, is the wind
Earth Observing System Aura satellite in 2004 (Levelt et al., 2006). The vector, and n is the unit vector orthogonal to the driving route. For the
OMI is an imaging spectrometer that covers the spectral range from measurements conducted in segments, the transportation and emission
270 nm to 500 nm. The OMI measured spectra are used to retrieve data flux of trace gases can be obtained by the following equation:
on O3, NO2, and other trace gases. The overpass time of the satellite is
about 13:45 LT. The USTC's OMI tropospheric NO2 product was com- Flux = VCDi vi sin i si (4)
pared with the mobile measurements. It is developed based on the
OMI's primary product through the DOAS fitting of OMI spectra (OMI where VCDi is the VCD of trace gases of segment i, vi is the wind speed
Level 1B VIS Global Radiances Data Product; https://disc.gsfc.nasa. corresponding to segment i, θi is the angle between the wind direction
gov/Aura/dataholdings/OMI/oml1brvg_v003.shtml). The Slant and driving direction of segment i, and ∆si is the distance between the
Column Densities (SCDs) of NO2 are separated to stratospheric and location of the two succeeding measurements i and i + 1.
tropospheric columns based on the local analysis of the stratospheric The wind field information was obtained through the WRF-Chem
field over unpolluted areas (Bucsela et al., 2013; Krotkov et al., 2017). simulation. First, the simulated high spatial (1 km × 1 km) and tem-
For the conversion of tropospheric OMI NO2 SCDs to VCDs, we re- poral (10 min) resolution 3-dimensional wind field data were weight-
calculated the AMF based on the NO2 and atmospheric profiles derived averaged by the WRF-Chem model-simulated trace gas profiles at dif-
from the Weather Research and Forecasting with Chemistry (WRF- ferent vertical heights (0–2 km). The horizontal wind field information
Chem) model simulations (Grell et al., 2005; Gao et al., 2016; Zhou was then interpolated to the position of the mobile measurements.
Fig. 2. Typical DOAS spectral fittings for (a) NO2 and (b) SO2. The spectrum was collected at an elevation of 30° at 13:59 LT on 15 January 2018. The red curves
show the reference absorption cross section scaled to the measured atmospheric spectrum (black curves) by DOAS fitting.
4
W. Tan, et al. Atmospheric Research 245 (2020) 105037
Fig. 3. The spatial distributions of NO2 and SO2 VCDs were measured on (a) 21–22 August 2017, (b) 15–17 January 2018 and (c) 09–10 February 2018 over NCP
region, China. The driving routes are marked by the dark thin lines, while the trace gases VCDs are indicated by colored thick lines.
3. Results and discussion of pollutants in winter (Ma et al., 2013; Hendrick et al., 2014). The SNR
of the instrument used in summer observations is relatively low in short
3.1. Spatial distribution of tropospheric NO2 and SO2 VCDs wave radiation where is the strong absorption wavelength range of SO2.
Meanwhile, the SO2 concentrations are low in summer. Therefore, the
Combined with the GPS logged position information, we obtained retrieved SO2 DSCDs were removed during the filtering of DOAS fit
the VCDs of different trace gases along the observation routes. Fig. 3 RMS and DSCD errors. The SO2 VCDs vary from 0.11 × 1015 molec
shows the spatial distributions of the NO2 and SO2 VCDs of the three cm−2 to 105.00 × 1015 molec cm−2 with an average of 29.47 × 1015
continuous long-distance observations crossing from the edge of the molec cm−2 in winter. These results were comparable to the previous
Yangtze River Delta region to the NCP region. In addition, the round- long-term ground-based MAX-DOAS measurements in this region, in
trip measurements in the center of the NCP region from Hengshui to which high values around 40–70 × 1015 molec cm−2 were observed in
Jinan passing through Dezhou are presented in Fig. 4. The dis- winter while the minimum of which around 20 × 1015 molec cm−2
continuous VCD plots are caused by the filter of the high SZA and RMS appeared in summer (Wang et al., 2014).
results. In the upper panel of Fig. 3 and Fig. 4, the spatial distribution of
It is obvious that the NO2 pollution in winter was more severe than NO2 VCDs shows the three high NO2 pollution regions located in
that in summer during this study. The NO2 VCDs vary from 3.54 × 1015 northern Anhui Province, central Shandong Province, and the route
molec cm−2 to 93.50 × 1015 molec cm−2 with a mean value of from Beijing passing by Baoding and ending in Hengshui. Furthermore,
35.97 × 1015 molec cm−2 in winter, while they range from the highest NO2 VCDs (i.e., > 81.30 × 1015 molec cm−2) were ob-
0.22 × 1015 molec cm−2 to 49.60 × 1015 molec cm−2 in summer with served on the way from Beijing to Hengshui on 17 January 2017. In
an average of 11.41 × 1015 molec cm−2. Higher levels of NO2 (~3 general, this is the heavily polluted area in the Beijing-Tianjin-Hebei
times) was obtained in winter, which can be explained by the more region of China, and the detailed explanation of this pollution process is
fossil fuel combustion for heating and the bad conditions for the spread provided in Section 3.3. In addition, the high value of NO2 VCDs were
5
W. Tan, et al. Atmospheric Research 245 (2020) 105037
Fig. 4. The spatial distributions of NO2 and SO2 VCDs were measured on (a) 24 January 2018 and (b) 27 January 2018 over center NCP region, China. The driving
routes are marked by the dark thin lines, while the trace gases VCDs are indicated by colored thick lines.
obtained usually when the car driving closer to the city center. The the mobile MAX-DOAS NO2 VCDs and USTC's OMI NO2 product, the
spatial distribution of SO2 VCDs during the campaign is reported in the mobile MAX-DOAS measured NO2 VCDs have been also compared with
bottom panel of Fig. 3 and Fig. 4, in which the three major heavily the NASA's standard OMI NO2 product for reference. The Pearson cor-
polluted regions are the same as the hotspots of NO2 distribution. relation coefficient (R) between the mobile measurement and USTC
However, the highest SO2 VCDs (i.e., > 88.20 × 1015 molec cm−2) product is 0.92 which is slightly higher than the Pearson correlation
were measured in Shandong Province centered at Jinan, Taian, and coefficient of 0.91 for the comparison with the NASA's standard pro-
Jining on 16 January 2018. As there are many power plants in these duct. In addition, the USTC product underestimated the tropospheric
areas, these high SO2 VCDs were expected (Xing et al., 2011). NO2 VCD by about 25% on average while this for NASA's standard
Furthermore, we compared the mobile observations and OMI sa- product increased to 40%. In general, the USTC product which uses the
tellite products to validate the spatial distribution of NO2 VCDs. First, NO2 profiles calculated from WRF-Chem simulations as a priori profile
the satellite data were filtered with the criteria of a relative error of less shows a better agreement with the mobile measurements. However, for
than 100% and the could fraction of less than 0.4. The daily spatial the mobile observations, which are more sensitive to pollution sources,
distributions of NO2 VCDs observed by the OMI and mobile MAX-DOAS there are wide ranges and variations in the mobile measurements while
were then plotted together in Fig. 5 for the days of 21 August 2017, 17, the satellite products with a slight variation, such as the observation on
24 January and 09 February 2018. It is obvious that the spatial dis- 21 August 2017. The mobile measurements range from 5.05 × 1015
tributions of the two datasets agreed well with each other, especially for molec cm−2 to 33.40 × 1015 molec cm−2 with a mean of
the OMI overpass time. However, the mobile observations also show (10.06 ± 6.52) × 1015 molec cm−2 while the satellite observations
higher sensitivity to the surface emission sources, such as large cities, vary from 5.25 × 1015 molec cm−2 to 9.42 × 1015 molec cm−2 with an
power plants, and refineries. Moreover, the pollution of large areas was average of (7.69 ± 1.32) × 1015 molec cm−2.
reported by both the OMI and mobile measurements on 17, 24 January
and 09 February 2018. 3.2. Observed NO2 and SO2 over different cities
For a better match in temporal and spatial coverage, the mobile
observed NO2 VCDs close to the OMI overpass time (12,00–14:00 LT) In order to characterize the air pollution in a large area, we com-
were averaged by the grid of 0.1° × 0.1°, while the OMI results were pared the NO2 and SO2 VCDs measured by mobile MAX-DOAS with the
averaged at the same grid. Fig. 6 shows the correlation analysis of these surface NO2 and SO2 concentrations measured by in-situ monitors over
4 days' observations in Fig. 5. Besides the correlation analysis between different cities of the NCP. The in-situ monitoring data is hourly
6
W. Tan, et al. Atmospheric Research 245 (2020) 105037
Fig. 5. Comparison of OMI and mobile MAX-DOAS measured NO2 VCDs on (a) 21 August 2017, (b)17 January 2018, (c) 24 January 2018, and (d) 09 February
2018.The mobile measurements were plotted overlap in the base map of OMI products. The routes of mobile measurements are indicated by the white lines, and the
position of the mobile measurements at 12:00 and 14:00 LT are marked by black triangles.
resolution (National Air Quality Historical Data, http://beijingair. from the city and the mobile MAX-DOAS observations are insensitivity
sinaapp.com/data/beijing/all/yyyymmdd/csv). The mobile MAX- to this far distance. In addition, the wind direction is also a strong in-
DOAS results were first hourly averaged. The value of the nearest fluencing factor, for the comparison between mobile MAX-DOAS ob-
segment (within 25 km) to the city center was then chosen to compare servations at the up-wind and down-wind areas of city may have sig-
with the in-situ observation results, which were the averages of mul- nificant difference. Meanwhile, here are also some explanations for the
tiple measurement sites in the cities during the same time period when difference in NO2 and SO2 comparison. First, the emission sources of
the mobile observation car passed by. As shown in Fig. 7, the con- NO2 and SO2 are quite different. The NO2 pollution is mainly from the
centrations of NO2 and SO2 in winter are obviously higher than those in traffic, industrial and power plant emissions. As the traffic emissions
summer, which were detected by both mobile MAX-DOAS and in-situ are widely distributed in the city, the in-situ monitor can observe the
observations. The time series of the NO2 levels of the two datasets are NO2 emissions sensitively. However, the SO2 pollution is major come
consistent well in Fig. 7(a), with a Pearson correlation coefficient (R) of from industrial and power plant emissions in the city. The SO2 pollution
0.64 (Fig. 8(a)). Different from the NO2 comparison, much more dif- is usually distributed near the factories and power plants, the spatial
ferences of the two datasets are presented for SO2 comparison in distribution of the SO2 pollution may not be as balanced as that of NO2
Fig. 7(b), and the Pearson correlation coefficient reduce to 0.42 pollution. Therefore, the mobile MAX-DOAS observation location far
(Fig. 8(b)). In addition, the reasonable explanation for the deviations away from the in-situ monitoring may lead to larger difference in the
between these two methods for some cases could be that the city center trends of SO2 VCDs and surface concentration. Second, the vertical
or the in-situ monitoring sites are far from the mobile observation distribution of NO2 and SO2 concentration is different. Usually, the NO2
routes. As shown in Fig. 7, the in-situ data report both peak values of pollution is mainly concentrated in the near surface, and the con-
NO2 and SO2 in Jinan on 16 January 2018, but the relatively lower centration decreases with the increase of height. Thus, the variation
values are obtained from the mobile MAX-DOAS observations. As the trend of VCDs is consistent well with the surface concentration detected
closest mobile measurement in this day is about 21 km (Fig. 7 (c)) far by in-situ monitor. However, the SO2 pollution are original from the
7
W. Tan, et al. Atmospheric Research 245 (2020) 105037
Fig. 7. Comparison of (a) NO2 and (b) SO2 results measured by mobile MAX-DOAS and in-situ measurement sites over different cities of NCP areas, China.
Meanwhile, the distances between the mobile MAX-DOAS observation and city center are presented in figure (c).
8
W. Tan, et al. Atmospheric Research 245 (2020) 105037
Fig. 8. The correlation analysis of mobile MAX-DOAS observations and in-situ measurements over different cities of NCP areas for (a) NO2 and (b) SO2, and the data
are accorded with the comparison in Fig. 7.
and wind field results corresponding to the highest NO2 VCD periods in here. Under the southeast wind, the pollutants spread in this area.
Fig. 10(a), (b), and (c) are presented in Fig. 10(d), (e), and (f), re- According to previous studies, this type of pollution process is very
spectively. common in this region (Xu et al., 2011; Fu et al., 2014).
In the first case shown in Fig. 10(a) and (d), owing to the large To quantify the transport of pollutants, the transportation flux es-
number of thermal power plants and factories located in northern timation method described in Section 2.3 was employed to calculate the
Jiangsu Province, local emissions are the major reason for the heavy flux. Fig. 11 shows the pollution case measured in northern Anhui
pollution. Under the easterly wind, the atmospheric pollutants from Province on 15 January 2018, where the colored line is the NO2 VCDs
northern Jiangsu Province were brought to northern Anhui Province. In along the mobile observation route. The hourly wind field information
addition, the pollutants were even spread to the areas adjacent to simulated by the WRF-Chem model during 15:00–16:00 LT is indicated
Henan Province. In the second case (Fig. 10(b) and (e)), the northerly by the black arrows, and the NO2 VCDs between the two black triangles
wind is dominant during the pollution period in central Shandong mark the positions measured during the same period. This shows that
Province. Under the action of northerly wind, the atmospheric pollu- the wind between the cities of Huainan (HN) and Huaibei (HB) is
tants in the Beijing-Tianjin-Hebei region may be transported to Shan- mainly from the east and almost vertical to the driving route. Therefore,
dong Province. Moreover, the low wind speed in central Shandong the segment (from 32.5° N to 34.0° N with a distance of about 200 km,
Province is unfavorable for the transport and spread of pollutants. In take about 1.5 h.) included in the black box was chosen for the esti-
the last case (Fig. 10(c) and (f)), the shape of the heavily polluted areas mation of the total transportation across the route. The total transpor-
of the Beijing-Tianjin-Hebei region looks like a strip. The northwest tation flux can be obtained through the integral along the observation
wind from northern Hebei Province and the southwest wind from route using Eq. (4) in Section 2.3.
Shandong Province simultaneously forced the pollutants to accumulate In addition, there is another method to acquire the total
Fig. 9. The ratios of SO2 to NO2, which were obtained by in-situ and mobile MAX-DOAS measurements.
9
W. Tan, et al. Atmospheric Research 245 (2020) 105037
Fig. 10. The spatiotemporal distributions of NO2 concentrations during the three typical pollution cases, which (a) is the NO2 VCDs measured by mobile MAX-DOAS
on 15 Jan 2018, (b) and (c) is same to (a) but for the observations on 16 and 17 Jan 2018, respectively, while the (d), (e) and (f) are the model simulated hourly
averaged surface NO2 concentration and wind field results corresponding to the highest polluted segment of (a), (b) and (c) (between the two short black lines),
respectively. The wind field information is indicated by black arrows.
transportation flux based on the simulated NO2 profile and wind in- mobile MAX-DOAS measured NO2 VCD and the model simulated wind
formation, as described in Eq. (5). The simulated NO2 VCDs at every information. As the wind fields at different height layers are different,
height layer are multiplied by the wind speed at the same grid and then some individual observations received a negative transportation value,
sum up from the surface to the height of 2 km. By obtaining the which meant that the pollutants are transported from west of the route
transportation of a single observation segment, the total transportation to east. On the contrary, positive values indicate that the pollutants are
flux is calculated by summing the transportation of all individual ob- transported in the opposite direction. Fig. 12(b) shows the transporta-
servations. tion of each individual observation segment and each height layer,
which are obtained by the model simulated NO2 VCDs and wind field
information. With more information on the vertical direction in the
Flux = cji vji sin ji hj si
i j (5) second method, it can be observed that the NO2 transportation is
mainly focused in the range from the surface to the height of 500 m,
where cji is the concentration of NO2, vji is the wind speed, θji is the and the highest transportation value is located at the height of about
angle between the wind direction and the driving direction, δhj is the 200 m. These vertical structural characteristics of NO2 transportation
height of layer j, and ∆si is the distance between the two successfully are significantly influenced by the vertical distribution characteristics
measured segments. of NO2 and wind field.
Consequently, both these two methods were used in the calculation The total transportation flux of this sectional segment is about
of the transportation flux of the pollution case in northern Anhui 7.12 kg s−1 and 7.09 kg s−1 using the measured and simulated NO2
Province. Fig. 12(a) shows the transportation of every individual ob- VCDs, respectively. The similar results of the two methods revealed that
servation segment between HN and HB, which are calculated by the
10
W. Tan, et al. Atmospheric Research 245 (2020) 105037
Fig. 12. The transportations of individual observation segment, which (a) is calculated by mobile MAX-DOAS measured NO2 VCDs and model simulated wind
information, while (b) is calculated by model simulated NO2 VCD and wind information.
11
W. Tan, et al. Atmospheric Research 245 (2020) 105037
Fig. 13. Comparison of the transportation results obtained by two different estimation methods, while (a) is the distance series plot and (b) is the correlation analysis
of the transportations obtained by these two different methods.
Table 3
The emission results of NO2 and SO2 from two different power plants.
References Observation Date Power plant Desulphurization Total installed capacity (MW) NO2 emissions (kg s−1) SO2 emissions (kg s−1)
This study 27 Jan. 2018 Dezhou, China Yes 2670 0.79 1.11
09 Feb. 2018 Hengshui, China Yes 1260 0.12 0.36
(Wu et al., 2017a) Oct. to Nov. 2011 Shijiazhuang, China Yes 600 0.107(±0.04) 0.078 (±0.036) (CEMS)a
(Frins et al., 2014) Mar. 2012 Montevideo, Uruguay No 385 11(±3) × 10−3 0.36 ± 0.12
a
Emission achieved from Continuous Emission Monitoring System (CEMS).
observation periods, the averaged NO2 and SO2 emission fluxes of this cross section scanning method cannot distinguish between the pollu-
power plant are 0.12 kg s−1 and 0.36 kg s−1, respectively. The total tants bring from upwind areas, power plant emissions, and even from
installed capacity of the Dezhou Power Plant is about 2.12 times higher other sources if there are considerable pollutants surrounding the
than that of the Hengshui Power Plant (Table 3), but the emission fluxes emission source, while the encircled observational method can obtain
of the Dezhou Power Plant are about 6.6 (NO2 emission fluxes) and 3.1 the net emission flux through subtracting the influx from the outflux.
(SO2 emission fluxes) times higher than those of the Hengshui Power Therefore, the emission flux of the Dezhou Power Plant may be slightly
Plant. Besides the difference in the total installed capacity, the different overestimated owing to the plume cross section scanning observation.
observation methods also can introduce some deviations. The plume Nevertheless, the results of the power plant emissions from the
12
W. Tan, et al. Atmospheric Research 245 (2020) 105037
mobile MAX-DOAS observations are reasonable and comparable to cm−2 during the winter observations. The comparison of the mobile
those of previous mobile DOAS studies listed in Table 3; for example, observations with the OMI tropospheric NO2 products show a good
the NO2 emission flux of the Hengshui Power Plant is close to that of a agreement with a high Pearson correlation coefficient (R) of 0.92. In
power plant of Shijiazhuang (Wu et al., 2017a). However, the total addition, the comparison of in-situ monitoring data and mobile mea-
installed capacity of the Hengshui Power Plant is about 2.1 times higher surements present a highly consistent variation trend, especially in the
than that of the power plant in Shijiazhuang. This means that the ef- comparison of NO2 concentrations. The ratios of SO2 to NO2 reveal the
fective denitrification measures in these years have reduced the NO2 emission characteristics of different cities.
emissions by nearly 46.6% for the Hengshui Power Plant. However, High concentrations of both NO2 and SO2 were detected in northern
owing to the different research methods, we obtain a different SO2 Anhui Province, central Shandong Province, and the Beijing-Tianjin-
emissions by the mobile MAX-DOAS observations in this study compare Hebei region during the winter observations in January and February
to those obtained in the Continuous Emission Monitoring System 2018. Combined with the WRF-Chem model simulated wind field in-
(CEMS) study of the power plant in Shijiazhuang. Moreover, the SO2 formation, the NO2 transportation from north Jiangsu Province to
emissions of the Hengshui Power Plant are consistent with those of the Anhui Province is quantified as 7.12 kg s−1. This meant that about
power plant in Montevideo, Uruguay (Frins et al., 2014). As the total 7.12 kg NO2 from north Jiangsu Province is transported to north Anhui
installed capacity of the Hengshui Power Plant is about 3.3 times higher Province under the action of easterly wind during the observation
than that of the power plant in Montevideo, the effective desulfuriza- period, thereby indicating the significant impacts of regional trans-
tion measures of the Hengshui Power Plant have reduced the SO2 portation of ambient air pollutants during the large-scale pollution
emissions by nearly 69.4%. Therefore, it is important to implement event.
denitrification and desulfurization processes before the emission of Finally, we estimated the NO2 and SO2 emissions of the Dezhou and
exhaust from power plants. In recent years, the Chinese government Hengshui power plants by scanning the plume cross section and en-
and relevant departments have formulated increasingly strict emission circling the emission source, respectively. The NO2 and SO2 emission
reduction measures, and desulfurization and denitrification equipment fluxes of the Dezhou Power Plant are 0.79 kg s−1 and 1.11 kg s−1,
is almost universal. Power plant emissions have now reached or even respectively, while the NO2 and SO2 emission fluxes of the Hengshui
exceeded the international standards (Schreifels et al., 2012; Xue et al., Power Plant are 0.12 kg s−1 and 0.36 kg s−1, respectively. These
2016) emissions are comparable with those of previous power plant studies,
and the comparison also highlights the importance of implementing
4. Conclusions desulfurization and denitrification processes before the emission of
exhaust from power plants.
Large-scale mobile MAX-DOAS observations were performed over
the NCP region in both summer and winter. Through the mobile ob- Data availability
servations, the distribution of NO2 and SO2 VCDs was obtained along
the measurement routes. In addition, we also quantitatively analyzed All data used in this study are available for scientific purposes upon
the pollutant transportation and emissions of different sources. request to the corresponding authors (shanshanwang@fudan.edu.cn
Through the spectral fitting and AMF conversion process, spatio- and chliu81@ustc.edu.cn).
temporal distribution information of NO2 and SO2 VCDs was obtained
over this region along with the GPS position. The measured average Author contributions
NO2 VCD is 35.97 × 1015 molec cm−2 in winter and 11.41 × 1015
molec cm−2 in summer, with a large variation between these two WT, CL, and SSW designed and implemented the research and wrote
seasons. In addition, the mean value of SO2 VCD is 29.47 × 1015 molec the paper; WT, and HRL carried out the mobile MAX-DOAS
13
W. Tan, et al. Atmospheric Research 245 (2020) 105037
14
W. Tan, et al. Atmospheric Research 245 (2020) 105037
15